[PATCH] Fix and add EXPORT_SYMBOL(filemap_write_and_wait)

This patch add EXPORT_SYMBOL(filemap_write_and_wait) and use it.

See mm/filemap.c:

And changes the filemap_write_and_wait() and filemap_write_and_wait_range().

Current filemap_write_and_wait() doesn't wait if filemap_fdatawrite()
returns error.  However, even if filemap_fdatawrite() returned an
error, it may have submitted the partially data pages to the device.
(e.g. in the case of -ENOSPC)

<quotation>
Andrew Morton writes,

If filemap_fdatawrite() returns an error, this might be due to some
I/O problem: dead disk, unplugged cable, etc.  Given the generally
crappy quality of the kernel's handling of such exceptions, there's a
good chance that the filemap_fdatawait() will get stuck in D state
forever.
</quotation>

So, this patch doesn't wait if filemap_fdatawrite() returns the -EIO.

Trond, could you please review the nfs part?  Especially I'm not sure,
nfs must use the "filemap_fdatawrite(inode->i_mapping) == 0", or not.

Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

fs/9p/vfs_dir.c

@@ -193,8 +193,7 @@ int v9fs_dir_release(struct inode *inode, struct file *filp)
 		fid->fid);
 	fidnum = fid->fid;
 
-	filemap_fdatawrite(inode->i_mapping);
+	filemap_write_and_wait(inode->i_mapping);
-	filemap_fdatawait(inode->i_mapping);
 
 	if (fidnum >= 0) {
 		dprintk(DEBUG_VFS, "fidopen: %d v9f->fid: %d\n", fid->fidopen,

fs/9p/vfs_file.c

@@ -165,8 +165,7 @@ static int v9fs_file_lock(struct file *filp, int cmd, struct file_lock *fl)
 		return -ENOLCK;
 
 	if ((IS_SETLK(cmd) || IS_SETLKW(cmd)) && fl->fl_type != F_UNLCK) {
-		filemap_fdatawrite(inode->i_mapping);
+		filemap_write_and_wait(inode->i_mapping);
-		filemap_fdatawait(inode->i_mapping);
 		invalidate_inode_pages(&inode->i_data);
 	}
 

fs/buffer.c

@@ -153,14 +153,8 @@ int sync_blockdev(struct block_device *bdev)
 {
 	int ret = 0;
 
-	if (bdev) {
+	if (bdev)
-		int err;
+		ret = filemap_write_and_wait(bdev->bd_inode->i_mapping);
-
-		ret = filemap_fdatawrite(bdev->bd_inode->i_mapping);
-		err = filemap_fdatawait(bdev->bd_inode->i_mapping);
-		if (!ret)
-			ret = err;
-	}
 	return ret;
 }
 EXPORT_SYMBOL(sync_blockdev);

fs/cifs/file.c

@@ -127,8 +127,7 @@ static inline int cifs_open_inode_helper(struct inode *inode, struct file *file,
 		if (file->f_dentry->d_inode->i_mapping) {
 		/* BB no need to lock inode until after invalidate
 		   since namei code should already have it locked? */
-			filemap_fdatawrite(file->f_dentry->d_inode->i_mapping);
+			filemap_write_and_wait(file->f_dentry->d_inode->i_mapping);
-			filemap_fdatawait(file->f_dentry->d_inode->i_mapping);
 		}
 		cFYI(1, ("invalidating remote inode since open detected it "
 			 "changed"));
@@ -419,8 +418,7 @@ static int cifs_reopen_file(struct inode *inode, struct file *file,
 		pCifsInode = CIFS_I(inode);
 		if (pCifsInode) {
 			if (can_flush) {
-				filemap_fdatawrite(inode->i_mapping);
+				filemap_write_and_wait(inode->i_mapping);
-				filemap_fdatawait(inode->i_mapping);
 			/* temporarily disable caching while we
 			   go to server to get inode info */
 				pCifsInode->clientCanCacheAll = FALSE;

fs/cifs/inode.c

@@ -1148,8 +1148,7 @@ int cifs_setattr(struct dentry *direntry, struct iattr *attrs)
 	/* BB check if we need to refresh inode from server now ? BB */
 
 	/* need to flush data before changing file size on server */
-	filemap_fdatawrite(direntry->d_inode->i_mapping);
+	filemap_write_and_wait(direntry->d_inode->i_mapping);
-	filemap_fdatawait(direntry->d_inode->i_mapping);
 
 	if (attrs->ia_valid & ATTR_SIZE) {
 		/* To avoid spurious oplock breaks from server, in the case of

fs/jfs/jfs_dmap.c

@@ -302,8 +302,7 @@ int dbSync(struct inode *ipbmap)
 	/*
 	 * write out dirty pages of bmap
 	 */
-	filemap_fdatawrite(ipbmap->i_mapping);
+	filemap_write_and_wait(ipbmap->i_mapping);
-	filemap_fdatawait(ipbmap->i_mapping);
 
 	diWriteSpecial(ipbmap, 0);
 

fs/jfs/jfs_imap.c

@@ -265,8 +265,7 @@ int diSync(struct inode *ipimap)
 	/*
 	 * write out dirty pages of imap
 	 */
-	filemap_fdatawrite(ipimap->i_mapping);
+	filemap_write_and_wait(ipimap->i_mapping);
-	filemap_fdatawait(ipimap->i_mapping);
 
 	diWriteSpecial(ipimap, 0);
 
@@ -565,8 +564,7 @@ void diFreeSpecial(struct inode *ip)
 		jfs_err("diFreeSpecial called with NULL ip!");
 		return;
 	}
-	filemap_fdatawrite(ip->i_mapping);
+	filemap_write_and_wait(ip->i_mapping);
-	filemap_fdatawait(ip->i_mapping);
 	truncate_inode_pages(ip->i_mapping, 0);
 	iput(ip);
 }

fs/jfs/jfs_txnmgr.c

@@ -1231,10 +1231,8 @@ int txCommit(tid_t tid,		/* transaction identifier */
 		 * when we don't need to worry about it at all.
 		 *
 		 * if ((!S_ISDIR(ip->i_mode))
-		 *    && (tblk->flag & COMMIT_DELETE) == 0) {
+		 *    && (tblk->flag & COMMIT_DELETE) == 0)
-		 *	filemap_fdatawrite(ip->i_mapping);
+		 *	filemap_write_and_wait(ip->i_mapping);
-		 *	filemap_fdatawait(ip->i_mapping);
-		 * }
 		 */
 
 		/*

fs/jfs/jfs_umount.c

@@ -108,8 +108,7 @@ int jfs_umount(struct super_block *sb)
 	 * Make sure all metadata makes it to disk before we mark
 	 * the superblock as clean
 	 */
-	filemap_fdatawrite(sbi->direct_inode->i_mapping);
+	filemap_write_and_wait(sbi->direct_inode->i_mapping);
-	filemap_fdatawait(sbi->direct_inode->i_mapping);
 
 	/*
 	 * ensure all file system file pages are propagated to their
@@ -161,8 +160,7 @@ int jfs_umount_rw(struct super_block *sb)
 	 * mark the superblock clean before everything is flushed to
 	 * disk.
 	 */
-	filemap_fdatawrite(sbi->direct_inode->i_mapping);
+	filemap_write_and_wait(sbi->direct_inode->i_mapping);
-	filemap_fdatawait(sbi->direct_inode->i_mapping);
 
 	updateSuper(sb, FM_CLEAN);
 

fs/jfs/resize.c

@@ -376,8 +376,7 @@ int jfs_extendfs(struct super_block *sb, s64 newLVSize, int newLogSize)
 	 * by txCommit();
 	 */
 	filemap_fdatawait(ipbmap->i_mapping);
-	filemap_fdatawrite(ipbmap->i_mapping);
+	filemap_write_and_wait(ipbmap->i_mapping);
-	filemap_fdatawait(ipbmap->i_mapping);
 	diWriteSpecial(ipbmap, 0);
 
 	newPage = nPages;	/* first new page number */

fs/jfs/super.c

@@ -502,8 +502,7 @@ static int jfs_fill_super(struct super_block *sb, void *data, int silent)
 		jfs_err("jfs_umount failed with return code %d", rc);
 	}
 out_mount_failed:
-	filemap_fdatawrite(sbi->direct_inode->i_mapping);
+	filemap_write_and_wait(sbi->direct_inode->i_mapping);
-	filemap_fdatawait(sbi->direct_inode->i_mapping);
 	truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
 	make_bad_inode(sbi->direct_inode);
 	iput(sbi->direct_inode);

fs/nfs/inode.c

@@ -644,10 +644,7 @@ int nfs_sync_mapping(struct address_space *mapping)
 	if (mapping->nrpages == 0)
 		return 0;
 	unmap_mapping_range(mapping, 0, 0, 0);
-	ret = filemap_fdatawrite(mapping);
+	ret = filemap_write_and_wait(mapping);
-	if (ret != 0)
-		goto out;
-	ret = filemap_fdatawait(mapping);
 	if (ret != 0)
 		goto out;
 	ret = nfs_wb_all(mapping->host);
@@ -864,8 +861,7 @@ nfs_setattr(struct dentry *dentry, struct iattr *attr)
 	nfs_begin_data_update(inode);
 	/* Write all dirty data if we're changing file permissions or size */
 	if ((attr->ia_valid & (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE)) != 0) {
-		if (filemap_fdatawrite(inode->i_mapping) == 0)
+		filemap_write_and_wait(inode->i_mapping);
-			filemap_fdatawait(inode->i_mapping);
 		nfs_wb_all(inode);
 	}
 	/*

fs/smbfs/file.c

@@ -374,8 +374,7 @@ smb_file_release(struct inode *inode, struct file * file)
 		/* We must flush any dirty pages now as we won't be able to
 		   write anything after close. mmap can trigger this.
 		   "openers" should perhaps include mmap'ers ... */
-		filemap_fdatawrite(inode->i_mapping);
+		filemap_write_and_wait(inode->i_mapping);
-		filemap_fdatawait(inode->i_mapping);
 		smb_close(inode);
 	}
 	unlock_kernel();

fs/smbfs/inode.c

@@ -697,8 +697,7 @@ smb_notify_change(struct dentry *dentry, struct iattr *attr)
 			DENTRY_PATH(dentry),
 			(long) inode->i_size, (long) attr->ia_size);
 
-		filemap_fdatawrite(inode->i_mapping);
+		filemap_write_and_wait(inode->i_mapping);
-		filemap_fdatawait(inode->i_mapping);
 
 		error = smb_open(dentry, O_WRONLY);
 		if (error)

fs/xfs/linux-2.6/xfs_fs_subr.c

@@ -79,8 +79,7 @@ fs_flushinval_pages(
 	struct inode	*ip = LINVFS_GET_IP(vp);
 
 	if (VN_CACHED(vp)) {
-		filemap_fdatawrite(ip->i_mapping);
+		filemap_write_and_wait(ip->i_mapping);
-		filemap_fdatawait(ip->i_mapping);
 
 		truncate_inode_pages(ip->i_mapping, first);
 	}

mm/filemap.c

@@ -343,30 +343,44 @@ EXPORT_SYMBOL(filemap_fdatawait);
 
 int filemap_write_and_wait(struct address_space *mapping)
 {
-	int retval = 0;
+	int err = 0;
 
 	if (mapping->nrpages) {
-		retval = filemap_fdatawrite(mapping);
+		err = filemap_fdatawrite(mapping);
-		if (retval == 0)
+		/*
-			retval = filemap_fdatawait(mapping);
+		 * Even if the above returned error, the pages may be
+		 * written partially (e.g. -ENOSPC), so we wait for it.
+		 * But the -EIO is special case, it may indicate the worst
+		 * thing (e.g. bug) happened, so we avoid waiting for it.
+		 */
+		if (err != -EIO) {
+			int err2 = filemap_fdatawait(mapping);
+			if (!err)
+				err = err2;
+		}
 	}
-	return retval;
+	return err;
 }
+EXPORT_SYMBOL(filemap_write_and_wait);
 
 int filemap_write_and_wait_range(struct address_space *mapping,
 				 loff_t lstart, loff_t lend)
 {
-	int retval = 0;
+	int err = 0;
 
 	if (mapping->nrpages) {
-		retval = __filemap_fdatawrite_range(mapping, lstart, lend,
+		err = __filemap_fdatawrite_range(mapping, lstart, lend,
-						    WB_SYNC_ALL);
+						 WB_SYNC_ALL);
-		if (retval == 0)
+		/* See comment of filemap_write_and_wait() */
-			retval = wait_on_page_writeback_range(mapping,
+		if (err != -EIO) {
-						    lstart >> PAGE_CACHE_SHIFT,
+			int err2 = wait_on_page_writeback_range(mapping,
-						    lend >> PAGE_CACHE_SHIFT);
+						lstart >> PAGE_CACHE_SHIFT,
+						lend >> PAGE_CACHE_SHIFT);
+			if (!err)
+				err = err2;
+		}
 	}
-	return retval;
+	return err;
 }
 
 /*